Skip to contents

Full-Form DCEA Tutorial

Source: vignettes/v03_full_dcea_tutorial.Rmd
v03_full_dcea_tutorial.Rmd

When to use full-form DCEA

Full-form DCEA is appropriate when subgroup-specific clinical evidence is available — for example, differential uptake by deprivation, differential survival benefit by SES, or trial subgroup data stratified by socioeconomic characteristics.

The inequality staircase

The five steps of the staircase define how patient-level benefit distributes across IMD groups:

  1. Disease prevalence — who gets the disease?
  2. Eligibility — who meets criteria for the treatment?
  3. Uptake / access — who actually receives it?
  4. Clinical effect — what QALY gain does each group achieve?
  5. Opportunity cost — what health is displaced in each group?

Worked example: NSCLC with subgroup evidence 

baseline <- get_baseline_health("england", "imd_quintile")

Subgroup-specific CEA inputs 

subgroup_data <- tibble::tibble(
  group       = 1:5,
  group_label = paste("IMD Q", 1:5),
  inc_qaly    = c(0.28, 0.36, 0.44, 0.51, 0.57),
  inc_cost    = c(13200, 12800, 12400, 12000, 11600),
  pop_share   = c(0.28, 0.24, 0.20, 0.16, 0.12)
)
subgroup_data
#> # A tibble: 5 × 5
#>   group group_label inc_qaly inc_cost pop_share
#>   <int> <chr>          <dbl>    <dbl>     <dbl>
#> 1     1 IMD Q 1         0.28    13200      0.28
#> 2     2 IMD Q 2         0.36    12800      0.24
#> 3     3 IMD Q 3         0.44    12400      0.2 
#> 4     4 IMD Q 4         0.51    12000      0.16
#> 5     5 IMD Q 5         0.57    11600      0.12

Differential uptake 

uptake <- c(0.58, 0.63, 0.68, 0.73, 0.77)

Run full-form DCEA 

result_full <- run_full_dcea(
  subgroup_cea_results       = subgroup_data,
  baseline_health            = baseline,
  wtp                        = 20000,
  opportunity_cost_threshold = 13000,
  uptake_by_group            = uptake
)
summary(result_full)
#> == Full-Form DCEA Result ==
#>   Net Health Benefit (equity-weighted): -0.1400 QALYs
#>   SII change: 0.2080
#>   Decision:   Lose-Lose (efficiency loss + equity loss)
#> 
#> -- Per-group results --
#> # A tibble: 5 × 5
#>   group_label inc_qaly_adj      nhb baseline_hale post_hale
#>   <chr>              <dbl>    <dbl>         <dbl>     <dbl>
#> 1 IMD Q 1            0.162 -0.220            52.1      51.7
#> 2 IMD Q 2            0.227 -0.176            56.3      55.9
#> 3 IMD Q 3            0.299 -0.122            59.8      59.5
#> 4 IMD Q 4            0.372 -0.0657           63.2      62.9
#> 5 IMD Q 5            0.439 -0.00770          66.8      66.6

Visualise

Inequality staircase 

sc_data <- build_staircase_data(
  group           = 1:5,
  group_labels    = paste("IMD Q", 1:5),
  prevalence      = c(0.08, 0.07, 0.06, 0.05, 0.04),
  eligibility     = c(0.70, 0.72, 0.74, 0.76, 0.78),
  uptake          = uptake,
  clinical_effect = subgroup_data$inc_qaly,
  opportunity_cost = subgroup_data$inc_cost / 13000
)